ppt - NKOS
Transcription
ppt - NKOS
Information Visualization and Semantic Search Xia Lin iSchool at Drexel College of Information Science and Technology Drexel University Philadelphia, Pennsylvania 19104 Overview ! Motivations and challenges of information visualization for semantic search ! Useful visualization for search ! Towards meaningful, useful and KOSbased visualization for search and discovery. 2 Information Visualization ! Definitions: – The use of computer-supported, interactive, visual representation of abstract data to amplify cognition. (Card, Mackinlay, and Shneiderman,1999) – Creation of visual approaches to conveying information in intuitive ways – Mapping of unstructured, linguistic data to structured visual space for easy understanding and discovery. 3 Motivations of Visualization ! Taking advantage of human cognitive capability for large amount of information processing, quickly and intuitively – Using visual representations to show large amount of information and enable visual inference. – Letting computer do mapping and analysis first and show only the most relevant associative information the user. – Moving information processing from the data/ linguistic level to the cognitive level. 4 Challenges of Visualization ! A picture is worth a thousand words if – it is meaningful ! The picture conveys semantic structures or relationships that the viewer can understand – It is trustful ! The structures and relationships on the picture match the semantic structures of the underlying data. – It is useful ! What users get from the picture will help them do something useful. 5 Meaningful? 6 Meaningful! Useful? 8 Useful! 9 More Examples 10 My Research Questions ! What functions are needed to make visualization useful? ! What characteristics of visualization displays will make the displays meaningful ? ! How to prepare the underlying data to make the visualization trustful and meaningful? 11 Useful Visualization Functions ! Overviews – Overview of large document space – Search result summary ! Dynamic Interactions – Network of documents, concepts, citations, – Neighborhood of the focused point ! Discovery – Visual analytics and discovery tools – Mapping of data to visual space 12 Washington Post – POTUS Tracker 13 Google+ ripples 14 Human Disease Network Graph 15 My Current Projects ! Visual Concept Explorer (VCE) – UMLS-based semantic vocabulary visualization and search ! Meaningful Concept Displays (MCD) – Getty-vocabulary based visual concept exploration and query expansion ! Visual Semantic Discovery Tools (VSDT) – Testbed: A million fulltext documents on neuroscience 16 Project 1- VCE2 17 Comparing to … 18 Project 2 - MCD ! IMLS funded project – To develop a Meaningful Concept Display (MCD) Appliance aiming to improve user’s searching, browsing, and learning experience with KOS and relevant content/collections. – To test the MCD appliance with ARTstor, Getty, and Indianapolis Museum of Arts (IMA) sites. – Collaborating with University of Buffalo, Getty, IMA, and ARTstor. 19 Demo 2 - VQE 20 Project 3 -- VSDT ! Visual Semantic Discovery Tools – Develop a triple-store knowledge repository to store the semantic entities and relationships extracted from full-text. – Design and implement visual analytics methods and interfaces to enable searching and knowledge discovery. 21 Demo 3 – Read with Annotation Graphical Summary 22 Summary: Visualize Semantics! ! Where do the semantics come from? – KOS + context + usage patterns – Annotation from text/context – citations/co-citations – Links/linked data ! How do we present the semantics to the user? – Meaningful – Trustful – Useful 23